You are here

Non-symmetrical DTD Iperf results

15 posts / 0 new
Last post
AB4YY
Non-symmetrical DTD Iperf results

On our 2.4 GHz network we have a new DTD setup with a Bullet on one side and a Rocket on the other.  The two nodes are roughly (guess) 400 feet away and the Rocket dish pointing mostly away from the Bullet's sector antenna.

The two devices are on ch. -2/10 but have different SSIDs.

Per WiFi Scans: Each of these two nodes hears the other at about -62dBm.

All of the above may be okay to live with although installation of a RF Armor kit is planned.


The problem is when I run IperfSpeed:

(Server -to- Client)

  • Bullet -to- Rocket = 36 Mbps
  • Rocket -to- Bullet  = 92 Mb/s

It doesn't seem right that there would be a 3:1 speed difference depending on which direction the testing is performed.

I did not personally set up the Netgear GS108E used for the DTD.  Both nodes have DHCP servers turned ON.

On a separate part of out mesh I tried IperfSpeed for a similarly DTD connected devices (two Rockets) and in each direction I see the same 90 Mbps.  So it makes me think something is wrong with our setup described above.  Any ideas?

Thanks, Mike ab4yy

k1ky
k1ky's picture
Non-symmetrical DTD Iperf results
My "guess" is that it probably has something to do with the fact that a Rocket is a MIMO (2 chain) device and the Bullet is single polarity (1 chain). Rocket numbers may be decoding on both chains? 
KE2N
KE2N's picture
no

this is a check through the LAN side and does not involve RF

AB4YY
I checked it again this
I checked it again this morning and the results are even worse but the same trend:
 
Bullet/Sector as Server -to- Rocket/Dish around 22 Mbps
Rocket/Dish as server -to- Bullet/Sector around 90 Mbps
 
Perhaps swapping the Netgear connections would give a clue.  This is not at my QTH so I can't try that at the moment.
 
It's still a mystery.
- Mike ab4yy
KE2N
KE2N's picture
horse power

The LAN connection on these units is 100 Mbps so getting 90 is pretty good.  I suspect that the one end of the iperf test does more work then the other end and the Bullet just does not have enough horsepower to support rates much above 25 Mbps (in addition to the other things it is doing - which will vary).

What you need to try is a Bullet-to-Bullet DtD test and see if that produces a symmetrical (but lower) speed result.

If true then this says using a Bullet at a central location with a lot of traffic might not be such a good idea.

Ken

KE2N
KE2N's picture
my test

so apparently the computing task for iperf is not all that onerous,   I added iperf to my airGrid node and did iperf across the Ubiquiti ToughSwitch in between over to the M5 and got 90 Mbps in both directions  (a symmetrical result).

It would seem most likely that you  have a bad cable or switch .... 

AB4YY
We replicated the devices
We replicated the devices (Bullet and Rocket) on another part of the network and do not see that non-symmetrical relation and the speeds are roughly 90 Mbps.  I even renamed the Rocket's SSID to something bogus.  That tells me the Bullet's horsepower is not the problem unless the subject one is defective which I doubt.
 
I think it may be more likely a wire pair in the ethernet transmissions from the Bullet to the Rocket that has an issue perhaps from a corroded connector outside.
 
I have requested the nodes be swapped at the Netgear switch.
 
- Mike ab4yy
KG6JEI
If you put a Support data
If you put a Support data file from the nodes we can look and see what it says about network errors.
AB4YY
There are no Internet

There are no Internet Gateways on our mesh but my connected computers (e.g. RPI, etc.) have Internet and I also have ssh access to the nodes.  Is there any other way to get the data?
- Mike

K6AH
K6AH's picture
A node's Support Data file

A link at the bottom of the node's Setup/Administration screen will download the Support Data file to your PC. 

Andre
 

AB4YY
Okay, I attached a .tar.gz

Okay, I attached a .tar.gz with the extension renamed to just .tgz
It opens okay with 7zip in Windows.
I couldn't see how to upload 2 support files.  :)
- Mike

KG6JEI
Thanks for the supportdata
Thanks for the supportdata files.

I'm not seeing the errors I would expect to see for a bad network cable, its still possible its the cause (and usually I would say its the  most probable scenario)  but it is odd to me that I'm not seeing network errors showing up.  If the cables can easily be swapped its worth testing a swapout to be sure, but I'll admit I'm a bit worried that may not actually solve the problem based on the lack of errors I'm seeing. 

I'll have to ponder more on the data.

*Tag Webmaster to comment about uploading multiple files in a single post*
WU2S
WU2S's picture
More support data file attachments possible
Thanks for the suggestion.
It is now possible to attach up to 4 support data files per post.
 
Support File Attachments: 
AB4YY
I am waiting for others at

I am waiting for others at the location(s) to try some swaps.  I can re-test then report back after testing.  I can easily test most any time but not physical site access to where the problem is.  - Mike

AB4YY
Issue closed!

Just to close this issue, there was a pair of DSL devices in the ethernet link for one of the nodes that had a flakey power supply.  Once the power supply was replaced, this entire issue went away.

Thanks for all the responses and suggestions with this issue.
- Mike ab4yy

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer